Security audit
Security checks for vulnerabilities and agentic risk
The skill is marked suspicious because the scan reports 1Password account listing outside its stated QA purpose and a privacy gap in live model evaluation; the evidence is incomplete.
Review the account-listing behavior and confirm it is needed for your use. Before live evaluations, check what data may be sent to model providers and use approved providers with sanitized inputs. The scan was partial because it could not resolve referenced paths, so treat the findings as incomplete evidence.
The supplied command queries 1Password account configuration rather than performing OpenClaw QA operations. No context establishes this account enumeration as a supporting implementation detail of the declared purpose. The command does not, by itself, retrieve vault secrets.
The live evaluation workflow sends scenario and conversation content to model backends without establishing a data-sensitivity check or approved-provider boundary. Persona content and workspace artifacts could contain confidential information that is then disclosed to candidate or judge providers; the instruction to preserve full transcripts also creates a secondary retention risk. Live model evaluation is a legitimate and explicit purpose here, so this is a conditional privacy-control gap rather than evidence of malicious exfiltration.
